Luca Soldaini 🎀
@soldaini.net
I like tokens! Lead for OLMo data at @ai2.bsky.social (Dolma 🍇) w @kylelo.bsky.social. Open source is fun 🤖☕️🍕🏳️🌈 Opinions are sampled from my own stochastic parrot
more at https://soldaini.net
more at https://soldaini.net
yea i was gonna link 🤣 rough guidelines I’ve heard for multilingual are around 600B+, which high level matches yuval’s findings.
October 21, 2025 at 5:50 PM
yea i was gonna link 🤣 rough guidelines I’ve heard for multilingual are around 600B+, which high level matches yuval’s findings.
text classification at scale, works great on 70TB of text
June 9, 2025 at 5:30 AM
text classification at scale, works great on 70TB of text
scales just fine to 70TB of text, supports subword embedding, someone made rust bindings 😌
June 9, 2025 at 5:29 AM
scales just fine to 70TB of text, supports subword embedding, someone made rust bindings 😌
no reason to switch just because the software is no longer updated. compile from scratch, works great!
June 9, 2025 at 5:28 AM
no reason to switch just because the software is no longer updated. compile from scratch, works great!
Reddit also has deals with OpenAI and GDM. Maybe negotiation stalled with Anthropic.
June 6, 2025 at 2:10 AM
Reddit also has deals with OpenAI and GDM. Maybe negotiation stalled with Anthropic.
they are a joy to type with our loud mechanical keyboards
June 3, 2025 at 6:52 PM
they are a joy to type with our loud mechanical keyboards
if soldering skills become critical i’m gonna be soon out of a job 😅
May 17, 2025 at 3:50 PM
if soldering skills become critical i’m gonna be soon out of a job 😅
MANGO SMOOTHIE
don’t forget da smoothie 🤤
don’t forget da smoothie 🤤
April 23, 2025 at 7:58 AM
MANGO SMOOTHIE
don’t forget da smoothie 🤤
don’t forget da smoothie 🤤
congrats!!! amazing news 🥰
March 27, 2025 at 3:29 AM
congrats!!! amazing news 🥰